Integration of a machine learning model into a decision support tool to predict absenteeism at work of prospective employees
Purpose - Inefficient hiring may result in lower productivity and higher training costs. Productivity losses caused by absenteeism at work cost U.S. employers billions of dollars each year. Also, employers typically spend a considerable amount of time managing employees who perform poorly. The purpose of this study is to develop a decision support tool to predict absenteeism among potential employees. Design/methodology/approach - We utilized a popular open-access dataset. In order to categorize absenteeism classes, the data have been preprocessed, and four methods of machine learning classification have been applied: Multinomial Logistic Regression (MLR), Support Vector Machines (SVM), Artificial Neural Networks (ANN), and Random Forests (RF). We selected the best model, based on several validation scores, and compared its performance against the existing model; we then integrated the best model into our proposed web-based for hiring managers. Findings - A web-based decision tool allows hiring managers to make more informed decisions before hiring a potential employee, thus reducing time, financial loss and reducing the probability of economic insolvency. Originality/value - In this paper, we propose a model that is trained based on attributes that can be collected during the hiring process. Furthermore, hiring managers may lack experience in machine learning or do not have the time to spend developing machine learning algorithms. Thus, we propose a web-based interactive tool that can be used without prior knowledge of machine learning algorithms.